Israel is looking to turn its AI expertise into something it can leverage beyond screens and software.
The country’s National AI Headquarters under the Prime Minister’s Office on Tuesday staged Israel’s first dedicated Physical AI conference in Tel Aviv, bringing together government officials, defense and technology companies, investors and global technology giants to examine the emerging intersection of AI, robotics and autonomous systems.
Held in partnership with B2B events producer People & Computers, the conference was primarily in Hebrew and presented the thesis that Israel’s Startup Nation identity could evolve into an “Edge-Up Nation,” in which Israeli AI is deployed not only in software but at the edge of the physical world.
The organizers described the event as an examination of the bridge between AI models and robotic and autonomous applications. Executives from Israeli robotics and autonomy companies including Innoviz, Beewise, Blue White Robotics, Pickommerce AI Robotics and LTX, as well as Shield AI’s Israel-linked leadership, attended the event alongside global technology companies including NVIDIA, Intel, AMD, Deloitte and Microsoft.
Speakers included Brig. Gen. (res.) Erez Eskel, head of the National Headquarters for Artificial Intelligence at the Prime Minister’s Office, and Lior Ben Horin, who leads NVIDIA’s Physical AI ecosystem work in EMEA.
A representative from the Prime Minister’s Office later confirmed to JNS that the conference served as the setting for Anduril Industries and Shield AI, two of the world’s largest defense-tech companies with a combined value of almost $74 billion, to discuss strategy and potential collaborations.
Guy Altagar, vice president of Physical AI at the Israeli National AI Directorate, also addressed the conference, together with Dov Moran, inventor of the USB flash drive and managing partner at Grove Ventures.
Israel’s leap into Physical AI
The focus reflected a broader shift in how Israel is thinking about artificial intelligence.
Generative AI largely produces information. Physical AI is designed to perceive the world, make decisions and act within it. That can mean an autonomous vehicle navigating a road, a robot operating in a warehouse, an agricultural machine working a field, or drones and other autonomous systems operating in environments where human intervention is limited.
Unlike cloud-based AI, edge AI allows machines to make decisions locally, which can reduce latency and enable systems to continue operating when connectivity is limited, rather than having to send data to centralized computing infrastructure for processing.
In Israel, the concept also has an obvious connection to the country’s defense sector, where drones, autonomous systems, sensors, computer vision and edge computing have been developed and deployed for years.
Oct. 7 and the subsequent war also accelerated demand for systems that can operate with less reliance on human operators, communications links or centralized computing. Israel’s recently approved National AI Strategy explicitly identifies “connecting software to the physical world” as one of the areas in which the country seeks to establish global leadership, building on capabilities developed in its defense industries.
That strategy is already moving from policy to implementation. In August, the Directorate published a Request for Information (RFI) to explore building a National Physical AI Testing Ground, a facility intended to allow companies and researchers to simulate complex real-world environments—from highways and hospitals to manufacturing plants and farms—to accelerate breakthroughs in robotics, edge computing and sensing technology.
Speaking to the media ahead of the conference, Altagar said the RFI has drawn more than 250 applicants, ranging from large international corporations to small startups.
“We invite a lot of different companies from a lot of different verticals, different fields of expertise, to participate and help us build this ‘kibbutz,’ this Israeli ecosystem,” he said. “We are very excited to … achieve the goals that Israel’s prime minister has developed for the national agenda: to start talking about AI not just as an application, [but] rather changing the way we live.”
Hila Khalil, the Directorate’s vice president of strategy, called the conference the “kickoff” for that Physical AI plan, with plans to form sector-specific communities and dedicated R&D centers in its wake.
The money question
The conference represented an Israeli attempt to identify where the country can build an advantage in the next phase of the AI race, following the initial explosion of generative AI and amid a growing global push toward robots and autonomous machines. But if the government’s pitch was about infrastructure and sovereignty, the event’s investor track explored how companies should go about asking investors to pay for it.
This national transition from an emphasis on software to Physical AI could challenge Israel’s tried-and-tested venture-capital model. Hardware requires more capital, longer development cycles and manufacturing capacity, whereas the traditional Israeli startup model has often been built around relatively capital-efficient software companies that can demonstrate rapid revenue growth.
That tension was explored in “Physical AI Needs Money: How Do You Deal With the Fear of It?,” a panel moderated by Dr. Assaf Horowitz, vice president of AI at the Israel Innovation Authority, that featured Orlie Sol Dahan Gruper of Robotics Factory, Dana Goldman Szekely of Greensoil Investments, 10D General Partner Itay Rand and E44 Ventures Managing Partner Adi Vagman.
The panel was mostly in Hebrew, but Ezra Gardner, CIO and co-founder of Varana Capital and CEO of Gesher Acquisitions Co., was among the panelists and spoke exclusively to JNS ahead of his appearance.
“All of the biggest companies in the public markets are actually hardware companies,” Gardner said, pointing to Amazon, Apple and Microsoft. “You ask yourself: Why is hardware not such a big VC thing, especially in Israel, but such a big thing in the public markets?”
His answer highlights the constraints of an early-stage investment cycle, in which building something like a semiconductor can require $100 million and five to 10 years before it generates revenue. Such a timeline can scare off software-focused VCs accustomed to funding a startup to an annual recurring revenue pitch within a couple of years.
Gardner’s advice to those founders was to stop pitching revenue they don’t have and instead find other proof points. He pointed to the country’s existing semiconductor and chip-design base, including Apple’s longstanding presence in Israel and NVIDIA’s plans to build a major campus in the north of the country.
Israel’s advantage
Sectors across Startup Nation are already developing many of the technologies that underpin Physical AI, from computer vision and sensors to autonomous vehicles, agricultural robotics and drones.
The Israel Innovation Authority counts more than 2,000 AI companies operating in the country, including 538 classified as deep-tech AI and roughly 95 specifically in robotics. In 2025, core AI companies captured 35% of all Israeli tech investment, part of a $14.6 billion year and representing a 30% jump over 2024.
Asked to define the term at the center of the conference, Gardner argued that “Physical AI” isn’t really new; rather, it is hardware and deep tech that have finally solved a specific old problem: giving machines something akin to a tactile sense. He predicted that the “Physical AI” label will eventually morph into the simpler term “robotics.”
“People are not going to need to say ‘Physical AI’ because robots are going to be everywhere,” he concluded. “It’s going to be obvious that it’s robotics, and they’re not going to say ‘AI’ because there isn’t going to be a dumb robot. They’re all going to be smart.”
